Towards free-standing MoS₂ nanosheet electrocatalysts supported and enhanced by N-doped CNT–graphene foam for hydrogen evolution reaction

Abstract

Large-scale free-standing porous carbon-based catalyst supports are critically needed for the hydrogen evolution reaction (HER) in view of their practical application. In this work, MoS₂ nanosheets were uniformly deposited onto N-doped carbon nanotube–graphene (N-CNT–G) hybrids, forming a three-dimensional (3D) free-standing architecture. The designed 3D hybrid materials exposed the MoS₂ nanosheet catalyst on the one-dimensional CNTs and there was facilitated ion transportation because of the porous graphene foam, and these 3D hybrid materials were used as electrocatalysts for HER directly without any transferring process. Moreover, N doping decreased the overpotential to 0.08 V, which increased the stability and promoted the catalytic activity, providing a facile route to design advanced high-performance HER catalysts towards a free-standing configuration.

---

1. Introduction

Hydrogen is a well-accepted clean alternative to the hydrocarbon fuels used today if it can be sustainably produced from water-splitting.^1,2 Electrochemical hydrogen evolution is a possible approach to meet this standard and has attracted increasing attention in recent years.^3,4 But one urgent need in the hydrogen evolution reaction (HER) is to find non-noble metal materials to replace the commonly used expensive and rare Pt catalyst.^5,6 Recently, molybdenum-based compounds have received more attention, such as Mo₂C,^7,8 MoP,^9,10 MoB¹¹ and so on. These Mo-based catalysts have very good HER activity. Research has also revealed that molybdenum sulfide (MoS₂) has outstanding electrocatalytic properties for the HER.^12–22 The active sites in MoS₂ are found to rely on its structure, these active sites come from the edge of the sheets in the crystalline platelets whereas they come from unsaturated sulfur atoms on the surface of amorphous MoS₂. However, the edge sites of MoS₂ are more catalytically active than the MoS₂ basal planes.^4,15,16 Both computational and experimental results confirmed that the HER activity stemmed from the sulfur edges of the MoS₂ plates.^16,18,22,25 In the water-splitting processes, the uncoordinated sulfur atoms denote active sites that bind hydrogen protons and release H₂ molecules.¹⁷ To improve the HER activity, exposing more of the surface and more edges of the MoS₂ using nanoparticles, nanosheets or porous structures allows the formation of a more active electrocatalyst.^17–19 To further improve its distribution and conductivity, the MoS₂ catalysts were used with carbon nanomaterials, such as carbon nanotubes (CNTs) and graphene as a support,^20–22 just like other types of electrocatalysts. Hence some well-established strategies, like using a CNT forest to increase the area-specific activity and applying N-doping to facilitate active species immobilization, were also suitable to produce advanced MoS₂-based electrocatalysts.^23,24 The recent study by Kim et al. has revealed that MoS_x/N-doped CNT (NCNT) forest hybrid catalysts exhibited the highest HER activity of any MoS₂-based catalyst ever reported.²⁵ Based on this progress and considering their future application, a further interest in this field is exploring MoS₂/heteroatom-doped CNT hybrid catalysts and especially free-standing architectures. Our previously developed CNT–graphene hybrid foams have been proven to form free-standing structures with high conductivity and large surface area, and rich in pores,^26,27 demonstrating the excellent characteristics of an ideal catalyst support. In this study, free-standing three-dimensional (3D) N-doped CNT–graphene (N-CNT–G) hybrid foams were fabricated and used as a support to grow MoS₂ nanosheets. The MoS₂/N-CNT–G electrocatalysts showed highly efficient electrocatalysis towards the HER and kept an unbroken 3D framework during the hydrogen evolution process due to its free-standing structure.

2. Experimental section

2.1. Synthesis of 3D non-doped and N-doped CNT–G hybrid foams

Non-doped CNT–G hybrid foams were synthesized as described in our previous work.^26,28 For the preparation of the N-CNT–G hybrid foams, pyridine instead of ethanol and benzene was used as the precursor to grow the N-doped graphene and CNTs. Here the remaining metal catalyst and Ni foam templates encapsulated in the graphene framework were removed by 3 mol L⁻¹ HCl solution at 90 °C for 10 h.

2.2. Synthesis of free-standing MoS₂/N-CNT–G hybrids

A piece of the N-CNT–G hybrid foam (∼5 mg) was carefully put into a Teflon-lined autoclave containing a preset amount of ammonium thiomolybdate ((NH₄)₂MoS₄) and 10 ml of N,N-dimethylformamide. The mass ratio of N-CNT–G to (NH₄)₂MoS₄ was controlled as 5 : 1, 1 : 3, 1 : 6 and 1 : 10. The reaction was carried out at 200 °C for 10 h and then cooled to room temperature naturally. Finally, the black MoS₂/N-CNT–G hybrid foam was brought out and dried in the air after being washed with ethanol and de-ionized water in turn. For comparison, MoS₂ was also deposited onto the CNT–G foam (termed as MoS₂/CNT–G) through a similar process.

2.3. Characterization

The products were characterized by scanning electron microscopy (SEM, Hitachi S-4800) and transmission electron microscopy (TEM, JEOL-JEM-2100F), X-ray diffraction (XRD, Philips X’pert Pro X-ray diffractometer with Cu Kα radiation of 1.5418 Å), X-ray photoelectron spectroscopy (XPS, PHI5000 VersaProbe) and Raman spectroscopy (Renishaw inVia Raman Microscope with an argon-ion laser at an excitation wavelength of 514 nm).

2.4. Electrochemical measurements

The electrochemical performance of the catalyst was evaluated by linear sweep voltammetry (LSV) and cyclic voltammetry (CV). All electrochemical measurements were conducted at room temperature in a three-electrode cell connected to a CHI 660C workstation (CH Instrument, Inc.). A slice of the free-standing MoS₂/N-CNT–G hybrid foam was bonded to a platinum wire to perform as the working electrode. A saturated calomel electrode and a graphite rod served as the reference and counter electrodes, respectively. The LSV was measured at a scan rate of 5 mV s⁻¹ in a 0.5 M H₂SO₄ solution.

3. Results and discussion

The N-doped graphene and CNTs were readily grown on Ni foams in turn by chemical vapor deposition. SEM images, shown in Fig. 1a and b, revealed a similar morphology to our previously reported non-doped graphene and CNT–graphene foams.²⁶ The presence of graphene and CNTs was revealed by the TEM images shown in Fig. 1c and S2.† As observed, the obtained multi-walled CNTs had diameters ranging from 20 to 60 nm. The content of the N dopant was about 1.06 at% according to the XPS analysis. It was mainly composed of pyridinic and pyrrolic N entities, which were depicted by the fitting peaks centered at 398.1 and 400.5 eV in Fig. 1d.^28,29 In order to obtain a freestanding N-CNT–G foam, the Ni foam templates encapsulated in the graphene framework were removed by HCl. As demonstrated in the SEM image of the N-CNT–G foam provided in Fig. S1,† the hybrid film still kept a free-standing feature in a regular foam structure, although some broken junctions were created when the Ni foam template was removed.

Fig. 1 (a) SEM image of the N-doped graphene/Ni foam; (b) SEM image of the N-CNT–G/Ni foam; (c) TEM image of the detached N-CNT and graphene sheets; (d) N 1s XPS narrow scan of the N-CNT–G/Ni foam.

Fig. 2a shows the typical SEM image of the MoS₂/N-CNT–G hybrid. The MoS₂/N-CNT–G hybrid has the same free-standing structure as the N-CNT–G foam at low magnification (Fig. S1a†). After MoS₂ deposition (Fig. 2b), the surface of the N-CNTs was uniformly coated by a layer of stacked nanosheets. The MoS₂/N-CNT composites inherited the one-dimensional morphology from the nanotubes while their diameter increased significantly. The morphology of the MoS₂/N-CNT–G was studied by TEM (Fig. 2c and S2†). The heavy MoS₂ coating made the N-CNTs indistinguishable under TEM observation. The MoS₂ nanosheets grew vertically out from the N-CNTs, and their thickness was estimated to be about 50–70 nm by comparing the diameter change between the N-CNTs (Fig. 1c) and MoS₂/N-CNT. It can be observed that each nanosheet was composed of 6–10 MoS₂ layers with a thickness of ∼8 nm (Fig. 2d). The interplanar distance of the lattice fringes was approximately 0.63 nm, corresponding to the (002) plane of MoS₂.^30–32 When deposited onto non-doped CNT–G, the MoS₂ also maintained a similar morphology (Fig. 2e and f) and structure to MoS₂/N-CNT. The MoS₂ layer on the CNTs, however, is thinner than that on the N-CNTs, suggesting that N doping is favourable for MoS₂ precursor adsorption and deposition.

Fig. 2 SEM (a and b) and TEM (c and d) images of the MoS₂/N-CNT–G hybrid. SEM (e) and TEM (f) images of the MoS₂/CNT–G hybrid. The inset in (d) is a locally enlarged TEM image of a MoS₂ nanosheet.

The XRD patterns of the MoS₂/N-CNT–G and MoS₂/CNT–G hybrids in Fig. 3a display two diffraction peaks at 31.9° and 56.2° that are attributed to the (100) and (110) planes of MoS₂.^33,34 The broad peaks indicated that the nanosized or even amorphous MoS₂ was synthesized by a hydrothermal method, which was very consistent with the TEM result (Fig. 2d). The other two peaks at 26.3° and 41.4° corresponded to the (002) and (100) planes of the graphite layers in the CNT walls. The characteristic features of the MoS₂ were also reflected in the Raman spectra (Fig. 3b), where the two bands at 384.2 and 408.7 cm⁻¹ were assigned to the in-plane (E¹_2g) and out-of-plane (A_1g) vibrational modes of the few-layer MoS₂.^35,36 In addition, the two spectra both gave two prominent peaks at 1340 and 1578 cm⁻¹, ascribed to D band and G band of the CNTs. The intensity ratio of the D band to the G band (I_D/I_G) in each sample was different. The N-CNT based sample showed a higher I_D/I_G because the N doping produced a lower graphitic crystalline structure.²⁸

Fig. 3 XRD patterns (a) and Raman spectra (b) of the MoS₂/CNT–G and MoS₂/N-CNT–G hybrids.

The composition and chemical state of the MoS₂ on the N-CNT–G foam was analyzed by XPS and is shown in Fig. 4. The Mo 3d XPS curve in Fig. 4a can be deconvoluted into two categories of Mo species. One is the dominant Mo(IV) that raised signal peaks of 3d_5/2 and 3d_3/2 at 228.8 and 231.9 eV, respectively. Another is Mo(VI), possibly from MoO₃ or MoS₃, as reflected by the fitted shoulder peaks in Fig. 4a. The fitted S 2s peaks (Fig. 4b) suggested that the S entities in our sample consisted of a majority of divalent sulfide ions (S²⁻) and a little bridging S₂²⁻, suggesting the existence of unsaturated sulfur atoms.^15,21,37 The mole ratio of S to Mo was measured to be 2.6, a little higher than the stoichiometric ratio of MoS₂, further supporting the existence some uncoordinated sulfur atoms. Hence the XPS analysis results given in Fig. 4a suggested that the obtained MoS_x should have been a mixture of dominant MoS₂ and a small amount of MoS₃ or MoO₃. Because the unsaturated sulfur atoms were inclined to gather at the edge sites of the as prepared MoS_x nanosheet, the S-rich structure would have given a negatively charged surface of molysulfide to facilitate proton adsorption and reduction to H₂ via electrostatic absorption of positively charged H⁺. The XPS spectra of Mo 3d and S 2p corresponding to the MoS₂/CNT–G and MoS₂/N-CNT–G (Fig. S4†) hybrids confirmed that the valence state of Mo and S within both compounds were +4 and −2, respectively. However, after N-doping, the binding energy of S 2p_1/2 and 2p_3/2 was slightly downshifted by 0.3 and 0.2 eV, respectively. As proposed by Kim et al.,²⁵ the S elements were principally connected with the N-CNTs. Hence the downshift of the S signal should have been caused by the protonation of lone-pair electrons from the N-doping sites.

Fig. 4 Mo 3d (a) and S 2p (b) XPS spectra of the MoS₂/N-CNT–G hybrid.

The proceeding results indicated that the free-standing MoS₂/N-CNT–G hybrids with S-rich MoS₂ nanosheets and conductive CNT–G frameworks were successfully prepared. The electrochemical HER tests were performed using a three-electrode setup in 0.5 M H₂SO₄ solution. Here only half of the free-standing MoS₂/N-CNT–G foam was immersed into the electrolyte (Fig. 5a) in order to avoid the effect of the Pt-wire conductor because Pt has a high activity for the HER. We chose the free-standing MoS₂/CNT–G hybrid and a commercial Pt/C catalyst (E-TEK) pasted on to a glassy carbon electrode as references to conduct parallel experiments. Fig. 5b shows the polarization curves of these three catalysts without correction for ohmic potential drop (iR) losses. The onset potential whose absolute value equals the overpotential (η) in the HER was close to zero for the Pt/C catalyst, suggesting it had high activity. The η of the MoS₂/N-CNT–G catalyst was 0.08 V which was a lot lower than the MoS₂/CNT–G catalyst (0.15 V). In addition, this η was also lower than some MoS₂ catalysts that were supported on pristine CNTs (0.09 V)²⁰ or graphene (0.1 V),²² suggesting that using a N-doping support could efficiently improve the activity of MoS₂ catalysts for the HER. Here N-doping provided an electron-rich surface for the N-CNTs, allowing the seamlessly anchored MoS₂ to create highly electronegative sites to catalyze the HER.^23,24,28 But our catalyst was slightly more inactive than the MoS_x/NCNT forest hybrid catalyst (η = 0.075 V) reported by Kim et al.,²⁵ probably due to their MoS_x having a higher S/Mo ratio or a more amorphous structure.

Fig. 5 (a) Photograph of the free-standing MoS₂/N-CNT–G catalyst tested in a three-electrode setup; (b) polarization curves and (c) Tafel plots of the Pt/C, MoS₂/CNT–G and MoS₂/N-CNT–G (1 : 3) catalysts; (d) polarization curves and (e) Tafel plots of the catalysts made from different initial precursor mass ratios of 5 : 1, 1 : 3, 1 : 6 and 1 : 10. (f) Cycling test of the MoS₂/N-CNT–G (1 : 3) catalyst for 1000 cycles.

The Tafel plots derived from the polarization curves were used to understand the catalytic process. The Tafel slope (b) in the linear regions of the Tafel plots (Fig. 5c) were calculated through fitting the Tafel equation (η = b log j + a, where j is the current density).³⁸ The calculated b was 30.2, 43.5, and 48.9 mV decade⁻¹ for the Pt/C, MoS₂/N-CNT–G and MoS₂/CNT–G catalysts. In general, there are three possible steps involved in the HER in acidic media, including the primary discharge step, the electrochemical desorption step and the recombination step (also named as the Volmer, Heyrovsky and Tafel reactions, respectively).³⁹ These three steps have a distinguishable b with a value of about 120, 40 and 30 mV decade⁻¹, respectively. As a result, we can roughly deduce which is the rate-limiting step according to the calculated b. In this case, the Pt/C catalyst gave a b of 30.2 mV decade⁻¹, suggesting that the recombination step was the rate-limiting step of the HER. For the MoS₂/N-CNT–G catalyst, the electrochemical desorption was the rate-limiting step because its b was 43.5 mV decade⁻¹. In addition, the HER catalyzed by it belonged to the Volmer–Heyrovsky mechanism, which was consistent with other types of MoS₂-based electrocatalysts.^21,25

With the aim of optimizing the catalyst, three control experiments were also conducted with different N-CNT–G : (NH₄)₂MoS₄ mass ratios of 5 : 1, 1 : 6 and 1 : 10. From the SEM images shown in Fig. 2b and S3,† it can be observed that the thickness of MoS₂ layer decorated on the N-CNTs increased with decreasing N-CNT–G : (NH₄)₂MoS₄ mass ratio. As a result, the catalyst prepared with the 5 : 1 mass ratio had the lowest MoS₂ loading while the one prepared with the 1 : 10 mass ratio had the highest loading. The compositions of these catalysts were analyzed by XPS as listed in Table S1.† The mole ratio of S to Mo for each sample varied from 2.4 to 2.6, similar to that of the MoS₂/N-CNT–G (1 : 3) catalyst (2.6). Their HER activities were tested in the same environment and are presented in Fig. 5d. The polarization curves indicated that these catalysts showed a lower HER activity than the MoS₂/N-CNT–G (1 : 3) catalyst, which was further supported by their Tafel slopes (Fig. 5e). Since all of the catalysts had a similar S/Mo mole ratio, the factors that could have possibly affected the activity were the loading and the structure of MoS₂ on the support. For example, the low loading of the catalyst prepared with the 5 : 1 mass ratio could not offer enough active sites while high loading for the catalyst (1 : 10) led to agglomeration of the MoS₂ nanoparticles. Both of them were harmful for catalyzing the HER. From the structural evolution of these supported MoS₂ catalysts, we could deduce that the nanosheet structure of the catalysts with the 1 : 3 mass ratio was beneficial for the exposure of active sites, and together with the proper loading amount, gave the highest activity for the HER.

Stability is also a significant property for HER catalysts. Our catalyst MoS₂/N-CNT–G (1 : 3) was assessed by the cyclic voltammetry method, cycling 1000 times from −0.7 to 0.3 V at 100 mV s⁻¹ (Fig. 5f). Compared with the initial polarization curve, the last one had almost unchanged features, indicating the strong interaction between the MoS₂ nanosheets and the N-CNT–G foam. Hence the MoS₂/N-CNT–G catalyst demonstrated good performance even in the free-standing configuration. After the electrochemical performance test, the used MoS₂/N-CNT–G (1 : 3) catalyst was characterized by SEM (Fig. S5†). The morphology was similar to the primary material, further supporting the high stability of this catalyst.

4. Conclusions

To prepare a free-standing high performance MoS₂-based electrocatalyst for the HER, CNT–graphene hybrid foams were used as a catalyst support and N-doping was applied to enhance the activity. With the assistance of N-doping, MoS₂ nanosheets were uniformly and tightly anchored on N-CNTs and formed a one-dimensional structure on the graphene foams. The as-prepared MoS₂/N-CNT–G catalyst showed a high activity and stability for the HER due to the additional electronegativity contributed by the N-doping. The good performance given by the free-standing configuration suggests that N-doped CNT–graphene hybrid foam is a potential catalyst support for the HER by optimizing active species.

Acknowledgements

This work is jointly supported by the Ministry of Education of China (no. IRT1148), NSFC (51402155), Priority Academic Program Development of Jiangsu Higher Education Institutions (YX03001), Jiangsu National Synergistic Innovation Center for Advanced Materials (SICAM), Synergistic Innovation Center for Organic Electronics and Information Displays, Jiangsu Provincial NSF (BK20141424).

Notes and references

1. M. S. Dresselhaus and I. L. Thomas, Nature, 2001, 414, 332–337.
2. J. A. Turner, Science, 2004, 305, 972–974.
3. L. Cheng, W. J. Huang, Q. F. Gong, C. H. Liu, Z. Liu, Y. G. Li and H. J. Dai, Angew. Chem., Int. Ed., 2014, 53, 7860–7863.
4. J. Kibsgaard, Z. B. Chen, B. N. Reinecke and T. F. Jaramillo, Nat. Mater., 2012, 11, 963–969.
5. H. B. Gray, Nat. Chem., 2009, 1, 7.
6. M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi, E. A. Santori and N. S. Lewis, Chem. Rev., 2010, 110, 6446–6473.
7. W. Cui, N. Y. Cheng, Q. Liu, C. J. Ge, A. M. Asiri and X. P. Sun, ACS Catal., 2014, 4, 2658–2661.
8. C. J. Ge, P. Jiang, W. Cui, Z. H. Pu, Z. C. Xing, A. M. Asiri, A. Y. Obaid, X. P. Sun and J. Tian, Electrochim. Acta, 2014, 134, 182–186.
9. Z. C. Xing, Q. Liu, A. M. Asiri and X. P. Sun, Adv. Mater., 2014, 26, 5702–5707.
10. W. Cui, Q. Liu, Z. C. Xing, A. M. Asiri, K. A. Alamry and X. P. Sun, Appl. Catal., 2015, 164, 144–150.
11. H. Vrubel and X. Hu, Angew. Chem., Int. Ed., 2012, 51, 12703–12706.
12. W. Cui, C. J. Ge, Z. C. Xing, A. M. Asiri and X. P. Sun, Electrochim. Acta, 2014, 137, 504–510.
13. Z. H. Pu, Q. Liu, A. M. Asirib, Y. L. Luo, X. P. Sun and Y. Q. He, Electrochim. Acta, 2015, 168, 133–138.
14. Z. H. Pu, Q. Liu, A. M. Asiri, A. Y. Obaid and X. P. Sun, J. Power Sources, 2014, 263, 181–185.
15. Y. Yang, H. L. Fei, G. D. Ruan, C. S. Xiang and J. M. Tour, Adv. Mater., 2014, 26, 8163–8168.
16. J. Kibsgaard, T. F. Jaramillo and F. Besenbacher, Nat. Mater., 2014, 6, 248–253.
17. B. Hinnemann, P. G. Moses, J. Bonde, K. P. Jørgensen, J. H. Nielsen, S. Horch, I. Chorkendorff and J. K. Nørskov, J. Am. Chem. Soc., 2005, 127, 5308–5309.
18. T. F. Jaramillo, K. P. Jørgensen, J. Bonde, J. H. Nielsen, S. Horch and I. Chorkendorff, Science, 2007, 317, 100–102.
19. J. Kibsgaard, Z. B. Chen, B. N. Reinecke and T. F. Jaramillo, Nat. Mater., 2012, 11, 963–969.
20. Y. Yan, X. M. Ge, Z. H. Liu, J. Y. Wang, J. M. Lee and X. Wang, Nanoscale, 2013, 5, 7768–7771.
21. H. Zhu, F. L. Lyu, M. L. Du, M. Zhang, Q. F. Wang, J. M. Yao and B.C. Guo, ACS Appl. Mater. Interfaces, 2014, 6, 22126–22137.
22. Y. G. Li, H. L. Wang, L. M. Xie, Y. Y. Liang, G. S. Hong and H. J. Dai, J. Am. Chem. Soc., 2011, 133, 7296–7299.
23. K. P. Gong, F. Du, Z. H. Xia, M. Durstock and L. M. Dai, Science, 2009, 323, 760–764.
24. S. J. Jiang, Y. W. Ma, G. Q. Jian, H. S. Tao, X. Z. Wang, Z. Hu and Y. Chen, Adv. Mater., 2009, 21, 4953–4956.
25. D. J. Li, U. N. Maiti, J. Lim, D. S. Choi, W. J. Lee, Y. Oh, G. Y. Lee and S. O. Kim, Nano Lett., 2014, 14, 1228–1233.
26. G. Y. Zhu, Z. He, J. Chen, J. Zhao, X. M. Feng, Y. W. Ma, Q. L. Fan, L. H. Wang and W. Huang, Nanoscale, 2014, 6, 1079–1085.
27. X. C. Dong, J. Chen, Y. W. Ma, J. Wang, M. B. Chan-Park, X. M. Liu, L. H. Wang and P. Chen, Chem. Commun., 2012, 48, 10660–10662.
28. Y. W. Ma, L. Y. Sun, W. Huang, L. R. Zhang, J. Zhao, Q. L. Fan and W. Huang, J. Phys. Chem. C, 2011, 115, 24592–24597.
29. L. F. Lai, J. R. Potts, D. Zhan, L. Wang, C. K. Poh, C. H. Tang, H. Gong, Z. X. Shen, J. Y. Lin and R. S. Ruoff, Energy Environ. Sci., 2012, 5, 7936–7942.
30. J. Deng, W. T. Yuan, P. J. Ren, Y. Wang, D. H. Deng, Z. Zhang and X. H. Bao, RSC Adv., 2014, 4, 34733–34738.
31. S. Han, Y. R. Zhao, Y. P. Tang, F. Z. Tan, Y. S. Huang, X. L. Feng and D. Q. Wu, Carbon, 2014, 81, 203–209.
32. K. Chang and W. X. Chen, ACS Nano, 2011, 5, 4720–4728.
33. S. Zhang, X. B. Yu, H. L. Yu, Y. J. Chen, P. Gao, C. Y. Li and C. L. Zhu, ACS Appl. Mater. Interfaces, 2014, 6, 21880–21885.
34. S. Qin, W. W. Lei, D. Liu and Y. Chen, Sci. Rep., 2014, 4, 7582.
35. J. F. Xie, J. J. Zhang, S. Li, F. Grote, X. D. Zhang, H. Zhang, R. X. Wang, Y. Lei, B. C. Pan and Y. Xie, J. Am. Chem. Soc., 2013, 135, 1788–17888.
36. K. M. Goodfellow, R. Beams, C. Chakraborty, L. Novotny and A. N. Vamivakas, Optica, 2014, 3, 149–152.
37. C. Y. Hu, S. Z. Zheng, C. J. Lian, F. Chen, T. W. Lu, Q. H. Hu, S. W. Duo, R. B. Zhang and C. Y. Guan, J. Mol. Catal. A: Chem., 2015, 396, 128–135.
38. B. E. Conway and B. V. Tilak, Electrochim. Acta, 2002, 47, 3571–3594.
39. W. C. Sheng, H. A. Gasteiger and Y. Shao-Horn, J. Electrochem. Soc., 2010, 157, B1529–B1536.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra05220c

---